Financial Technology and AI

Module 0

Author
Affiliation

Prof. Matthew G. Son

University of South Florida

Financial Technology

What is FinTech?

  • The verbatim definition: Financial Technology

FinTech

  • The buzzword FinTech started to appear around 2015

  • Google Search Trend

  • FinTech has its own specific connotation regarding:

    • Business model

    • Area of Finance

    • Technologies applied in Finance

FinTech by Business Model

  • FinTech companies expands the customer base by:

    • Democratizing technology

      • through use of mobile apps (e.g. Robinhood, Chime)
    • Large cost-reduction with innovative/streamlined business model

      • through automated models based on big data
      • can have large customer base compared to size of the firm
  • B2B, B2C oriented -> P2P, Platform

FinTech by Financial Sector

  • Payment systems, Brokerage/Exchange services (Platform)

    • Paypal (Venmo), Zelle

    • Robinhood, Coinbase

  • Investments and Portfolio management

    • Robo-advising (Portfolio management)

    • High-Frequency, Algorithmic Trading (Mutual/Hedge Funds)

  • Credits and Loans (Banking)

    • Crowdfunding, P2P lending (LendingClub)

FinTech by Technology

The financial companies that use innovative technologies

  • Big data
    • by volume & velocity: large, high frequency data
    • by variety: image, sound, text data
  • Machine Learning
  • Blockchains/ Cryptocurrencies
  • Internet of Things
  • Virtual Reality / Metaverse

Peek at FinTech Business models

Sample start-up companies and their business models:

  1. Numerai
    • Hedge fund leveraging crowdsourced models and big data
  2. Zest AI
    • Credit risk analysis and automated underwriting with alternative data
  3. Niural
    • global team hiring, tax compliance and payments

TechFin

FinTech in the Globe

FinTech in Tampa & St. Pete

Our strategic partnership program to grow network of innovators in FinTech world:

  • https://www.tampabaywave.org/

Introduction to AI

AI, ML, DL

Artificial Intelligence

Artificial Intelligence (AI) : intelligent machine

Those machines are called intelligent as:

  • They have decision-making capabilities

  • like human beings (i.e., “smart”)

Chess and AI

Deep Blue vs. Gary Kasparov (1997)

  • IBM’s Deep Blue defeats world chess champion Kasparov.
  • Marked a milestone in AI with rule-based systems and brute-force calculations.

Programming and ML

Machine Learning

Machine Learning (ML): A subset of AI

  • A technology to allow computer to learn the patterns and rules
  • Without specific instructions

Algorithms make the computer to learn

  • Some from Statistics, others from Computer science

  • Neural network (Deep Learning) one of such algorithm

Programming and ML

Go (Baduk) and AI

AlphaGo vs. Lee Sedol (2016)

  • AlphaGo defeats Go legend Lee.
  • AlphaGo Zero (2017): Self-learning reinforcement model that surpassed human expertise.

Tesla FSD

Previous versions (V1 ~ 11)

  • Programmed approach
  • Used hundreds of thousands of hard-coded C++
  • Rigid, robotic and jerky driving style

Since 2024 (V12 ~)

  • Machine Learning (Deep Neural network)
  • Used “safe” human driver data for training
  • Improved, natural, human-like driving

AI and ML examples

AI, not ML:

  • Automation
  • Timers (IoT)
  • Grammar/syntax checker (earlier)

ML:

  • Chatbots (Gen-AI)
  • Image classification
  • Anomaly detection
  • Self-driving

Finance example: Loans

If loan officers applied certain known rules for loan decisions:

  • we ask what rules they applied (should be clear-cut)

  • this process can be automated by set of instructions (programming)

  • AI (non-ML)

Finance example: Loans (Cont’d)

If we do not know what rules are used, but the outcomes:

  • Subjective and fuzzy rules, not always the same rules applied

  • We could use ML to learn the rules using data

  • ML will make automated decisions based on the data patterns

  • AI (ML)

Limits of ML

Complexity-Interpretability Tradeoff

Interpretability Issues

Black Box

  • Financial data is inherently complex and interdependant

  • Hard to track why a specific decision (e.g., loan approval) was made

  • Financial regulators and institutions require clear reasoning behind decisions

  • Even harder for complex ML algorithms

AI: The 4th Industrial Revolution

How will it change the financial world?

The Future That Arrived Early

Some insights from how AI shaped the world of Go

  • One of the first shocked field that was disrupted by AI (DL)
  • A game of longest human history (BC 2200)
  • A perfect information game, solvable in theory, not in practice

AI Disruptions

Corner opening moves (joseki)

  • AI challenged some traditional, long-accepted patterns (found suboptimal)

  • No longer played (deprecated)

  • Play focus: imaginative to calculative

    • Convergence of player’s style (meta-game)

Limits of AI Disruption

  • Humanity aspect

    • Nobody wants to watch machine games
    • Charm of imperfection
  • Simplicity (interpretability) vs accuracy tradeoff

    • Extreme complexity: too hard to understand
    • Fallback to understanable (safe) move

AI Empowering Human

AI Empowering Human (Cont’d)

Dramatic improvement in average Human Go skills

  • Best in 2012 vs 2016 vs 2020: who wins?

  • AI discovers, confirms and strengthens human theory

New AI-based josekis are found and adopted

  • Players learn from AI’s playstyle

  • Commentators use AI for win rate

Human Learning

AI is less effective for teaching the game for beginner-intermediates:

  • It can overwhelm novices
  • Can’t understand AI’s complex level
  • Often fails to deliver foundational insights
  • Ineffiective intuitions

AI becomes more important at high skill levels:

  • Far easiaer access to sophisticated level knowledge
  • For advanced learners approaching mastery

Insights for Finance

In comparison, the field of Finance is

  • A game of incomplete / imperfect information.

Insights for Finance

  1. Financial Accountability:
  • High scrutiny fields: like banking, auditing, and compliance that need transparency will less likely be impacted

  • Innovation-friendly: asset management, quantitative trading will more likely impacted

Insights for Finance

  1. Homogenized investment strategy:
  • Less distinctive portfolio styles / Portfolio herding

  • Potentially leading to higher systemic risks

Insights for Finance in the age of AI

  1. Soft Information Matters:
  • Qualitative (big) data will be crucial for complementing hard-data

  • A competitive edge for firms that efficiently process alternative data

  • Appreciation of Proprietary data value

  • Increased importance of privacy & cybersecurity